These are the proceedings of the third Linguistic and Cognitive Approaches to Dialog Agents (LaCATODA 2018), which was held on July 13, 2018 in Stockholm, Sweden. The workshop was a part of the 27th International Joint Conference on Artificial Intelligence (IJCAI 2018) colocated with ECAI 2018, the 23rd European Conference on Artificial Intelligence. Bothe conferences were held under the umbrella of Federated AI Meeting (FAIM 2018), the biggest AI joint conference so far. Fourteen papers were submitted to the workshop of which eight was accepted to be presented during the workshop. The submissions were reviewed by three (at least two) members of the program committee. This workshop was the fourth edition of the LaCATODA series. The two first editions were organized as symposia of the AISB/IACAP conferences in United Kingdom: the first one in 2010 in Leicester, and the second one in Birmingham. The third edition was co-located with the prestigious IJCAI conference in Melbourne, 2017. We would like to thank all authors who submitted papers, and the program committee members for their efforts.
Rafal Rzepka, Jordi Vallverdu and Andre Wlodarczyk Copyright © 2018 for the individual papers by the papers’ authors. Copying permitted for private and academic purposes. Re-publication of material from this volume requires permission by the copyright owners. This volume is published and copyrighted by its editors. Program Chairs / Organizers / Editors:
Language Media Laboratory Graduate School of Information Science and Technology Hokkaido University, Kita-ku Kita 14 Nishi 9, 060-0814 Sapporo, Japan Jordi Vallverdú - jordi.vallverdu@uab.cat Autonomous University of Barcelona, Spain Program Committee Improving Goal-Oriented Visual Dialog Agents via Advanced Recurrent Nets with Tempered Policy Gradient ……………………………………………………………………….…………… 1 Rui Zhao and Volker Tresp Detecting Location-Indicating Phrases in User Utterances for Chat-Oriented Dialogue Systems ……………………………………………………………………………………………… 8 Hiromi Narimatsu, Hiroaki Sugiyama, Masahiro Mizukami Efficient Purely Convolutional Text Encoding.…………………………………………………… 14 Szymon Malik, Adrian Lancucki, Jan Chorowski Dialogue Modeling Via Hash Functions …………………………………………………………… 24 Sahil Garg, Guillermo Cecchi, Irina Rish, Shuyang Gao, Greg Ver Steeg, Sarik Ghazarian, Palash Goyal and Aram Galstyan Towards a structured evaluation of improv-bots: Improvisational theatre as a non-goal-driven dialogue system……………………………………………………………….……… 37 Maria Skeppstedt and Magnus Ahltorp Refinement of utterance database and concatenation of utterances for enhancing system utterances in chat-oriented dialogue system…………………………………..…… 44 Yuiko Tsunomori, Ryuichiro Higashinaka and Takeshi Yoshimura Event Data Collection for Recent Personal Questions……………………………………… 52 Masahiro Mizukami, Hiroaki Sugiyama and Hiromi Narimatsu